Fix fp16/q4f16 dtype test failures on Node.js 20+ #1481
+22
−4
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
onnxruntime-common uses Float16Array when available (Node 20+), but onnxruntime-node's native binding only accepts Uint16Array for float16 tensors. This causes dtype tests to fail locally with: TypeError: Tensor.data must be a typed array (4) for float16 tensors, but got typed array (0). Add a workaround that hides Float16Array before onnxruntime loads, forcing it to use Uint16Array instead. Upstream PR to add Float16Array support: microsoft/onnxruntime#26742
Collaborator
|
Hi 👋 thanks for the PR. Preferably, we'd be able to fix this for normal running, not just unit tests. Any ideas? It seems like any call to |
Collaborator
|
I did a bit of digging and #1482 seems to fix it. How does that look? |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Problem
Preparing a different PR I noticed that running tests locally on Node.js 20+ causes the Llama fp16/q4f16 dtype tests to fail:
Cause
onnxruntime-commondetectsFloat16Array(available in Node 20+) and uses it for float16 tensors. However,onnxruntime-node's native binding was written beforeFloat16Arrayexisted and only acceptsUint16Array.Fix
This PR adds a workaround that hides
Float16Arraybefore onnxruntime loads, forcing it to useUint16Arrayinstead.There's an upstream PR to add
Float16Arraysupport to onnxruntime-node (microsoft/onnxruntime#26742), but it hasn't merged yet (and onnxruntime PRs can take many months to merge). Once it does, this workaround can be removed.Test plan
npm testpasses locally on Node.js 20+